Abstract: Combining with advantages of neural network method and genetic algorithm, a method to optimize machining process parameters was proposed for thin-walled workpieces based on back propagation neural network(BPNN). The data gained from Taguchi experiments were applied to train in BPNN so as to generate the S/N ratio predictor and quality predictor. By maximizing the S/N ratio, variation of milling processes was minimized, and the optimal process parameter combinations were found. Through numerical simulation and machining experiments, effectiveness of the proposed method in optimization of milling process parameters of thin-walled workpieces was validated.

Key words: thin-walled workpiece, taguchi method, genetic algorithm, processing parameter optimization

摘要： 结合神经网络法和遗传算法的优点，提出了一种以倒传递神经网络法为基础的加工工艺参数优化方法，对薄壁件铣削加工工艺参数进行优化。将田口实验所得数据经倒传递神经网络进行训练与测试，来建立薄壁件铣削加工的信噪比预测器，并通过最大化信噪比，将铣削过程变异降至最低，进而找出最佳加工工艺参数组合。通过数值模拟与加工实验，验证了所提方法在薄壁件铣削加工工艺参数优化中的有效性。

关键词: 薄壁件, 田口法, 遗传算法, 工艺参数优化